To have a clear picture of monetary conditions, one must be able to say whether money is tight or loose. The question is less clear than one might have thought—here in late 2025, after a period of rising short-term rates, followed now by one rate cut, is liquidity scarce or not? After a long period of easy conditions, the zeroing out of one important balance sheet quantity makes a clear break on the path from loose to tight money.

The rise and fall of ON RRP

At the end of 2022, balances at the US balances at the US Federal Reserve’s overnight reverse repo (ON RRP) deposit facility stood just north of $2.5 trillion. As of yesterday (2 Oct 2025) the figure was $8 billion. Some day soon it may touch zero. The ON RRP facility is significant for its role in accommodating the Fed’s asset purchases at the height of the COVID-19 pandemic. This picture of the US central bank’s balance sheet points us in the right direction:

The ON RRP deposit facility is a liability for the Fed, along with currency, the reserve deposits of US commercial banks and the US government’s deposits (the Treasury General Account or TGA). Following the outbreak of the pandemic in early 2020, the Fed followed its playbook from the 2008 crisis, launching a quantitative easing program that would eventually bring its balance sheet to more than $9 trillion. On the asset side, expansion took the form of purchases of Treasury securities and agency mortgage-backed securities.

To accommodate the asset purchases, the Fed had to allow expansion on the liability side as well. At various points both reserves and Treasury deposits grew, but by far the most interesting entry was ON RRP. The facility had been tested from 2014 to 2018, but those tests reached only about $100-200 billion. The Fed’s COVID-era asset purchases flooded the system with newly created money, and for a while ON RRP, meant to be a worst case, was the only option, particularly for money market mutual funds. It is not a stretch to say that the US central bank had to create an entirely new category of high-powered money to allow its balance sheet to double over the course of two years.

Starting in 2022, the Fed switched to contraction, and since 2023 that contraction has mostly been balanced on the liability side by reductions in the ON RRP facility. It has been contracting ever since, and is now almost closed out.

I guess that means everything’s back to normal

At the very least, it is a milestone, the end of an instrument that played a key part in the operation of the global dollar over the last five years. From an institutional point of view, it marks an end to the Fed’s five-year run of providing deposit services to money market funds. Back to the pre-pandemic normal, in that sense, where central money is held by the government and the commercial banks.

Or to put it another way—contraction so far has been absorbed by ON RRP; that pool is now exhausted. Any further contraction will necessarily be different.

Previously on Soon Parted

• The post-pandemic Fed: early hints

• The Fed in the repo market: The ascent of market-based finance

• Money is not yet tight, rate hikes notwithstanding

Besides the connection to the auto industry, the two collapses have little in common. Tricolor integrated auto sales, collections and collateral management to support profitable issuance of commercial paper backed by its auto loans; First Brands issued debt to finance acquisitions while optimizing cash flow with extensive supply-chain financing.

Stability in the financial system doesn’t necessarily mean that bankruptcy never happens—it means that failures should not be transmitted upward in the financial hierarchy. Either of these bankruptcies, Tricolor or First Brands, could have passed without prompting wider concern about stability. In the event, however, the two have come simultaneously, which is enough to make credit markets jumpy.

Tricolor Holdings

Tricolor Holdings offered a modern, integrated take on the subprime auto lending market. Targeting borrowers with cash incomes and limited credit history, the company is best understood as a retail loan originator with an integrated collateral-management business. Tricolor’s loan terms typically demanded down payments of 11% and interest rates of 17%, based on a biweekly payment schedule.

The company paired these loans with collateral repossession practices that allowed it to make the most of the steep upfront payments: after default, vehicles could be quickly brought back to the dealership, cleaned up and put up for sale again. Each time the car is sold, the company pockets 11% of its value, paid in cash up front. As long as there are enough borrowers, it is better for Tricolor if they default than if they pay.

This business model allowed Tricolor to issue $1.4 billion in auto loan securitizations. Lenders and bond holders include Fifth Third, JPMorgan, Barclays, part of the Massachusetts state pension fund, community bank Triumph Financial, bond holders Charles Schwab, Pimco, and Capital Group. The company ran into trouble in recent weeks, due at least in part to high levels of defaults driven by the deportations of its borrowers. To cover the cash flow, it seems that Tricolor may have started double-counting collateral. Creditors got wind and opened an inquiry in August, and the company declared bankruptcy two weeks later.

First Brands

This week, auto parts maker First Brands Group also filed for Chapter 11 protection. The company grew rapidly over the last few years in a debt-funded expansion. It seems to have gotten into trouble with its trade financing practices, recorded in so-called off-balance-sheet instruments. Just because First Brands kept the arrangements off the books, of course, doesn’t mean that we have to do the same.

FBG borrowed against its own invoices, as shown in the balance sheets below. Having sold its product, the company could pledge its unpaid invoices, as well as its inventory, to investors for cash. When the proceeds of the sale are eventually collected, the funds can be used to pay the loans, with a premium to the lender for supplying liquidity. This is a standard commercial credit arrangement (called factoring when secured by invoices, but equivalent to any other kind of secured lending). It is normally a simple way to bridge the gap between delivery and receipt of payment:

At the same time, FBG borrowed to cover its own purchase of inputs. These reverse factoring arrangements achieve a similar effect (smoothing over payment delays) as invoice financing, but are initiated by the buyer instead of the seller:

When it failed this week, First Brands seems to have had as much as $10 billion in debt and off-balance sheet financing outstanding. The company’s creditors include the investment bank Jefferies and UBS’s O’Connor hedge fund. Some eagle-eyed investors raised concerns over an August refinancing deal, and the added scrutiny quickly revealed a significant cash shortfall.

It’s the sudden stop at the end

Both Tricolor and First Brands came apart in a matter of weeks. Whatever the underlying problems, it is always illiquidity that brings on such a rapid end. Both of these companies met their end when they finally ran out of cash.

The differences in circumstances are therefore either deeply reassuring or deeply worrying, depending on your interpretation. Perhaps the two failures are unrelated, simply a bad roll of the dice in the business-as-usual flow of creative destruction. Then a few lenders will get burned, lessons will be learned, or maybe not, but either way it doesn’t mean much about what comes next. Alternatively, the synchronous failure of two unrelated companies could be a sign of trouble at a deeper level. If they indicate tighter money, a general flight towards cash, then there may be more turbulent financial conditions to come.

Previously on Soon Parted

• Archegos

• Greensill

• Plugging the leaks: three refinancing channels, so far

• Stablecoins are banks

DVP repo has printed above the fed funds upper bound and below the fed funds lower bound since the September 2024 rate cut: money is tighter, but also less stable—reminiscent of 2019 and quite unlike 2023.

Previously:

• Basis

Repo drain runs dry; issuance now less ample. QT is ending.

Previously

Money is not yet tight

Subscribe now

Treasury futures: a trillion open int’rest. What couldn’t go wrong?

Previously

• Cash–futures basis: The missing T accounts

Soon Parted is back, its infrastructure rebuilt. Expect cryptic posts.

Previously

• Cash–futures basis: The missing T accounts

Five years on, the 2019 repo crunch is eclipsed by the pandemic, both in memory and in lingering macroeconomic effects. But unlike the pandemic, the crunch was brought on—at least in part—by contractionary policy. Policymakers are keen to show they have learned from the experience.

This graph shows the Fed’s balance sheet from before the 2008 crisis all the way to the present. Assets are above, liabilities below. Vertical lines mark the important turning points in the central bank’s balance-sheet policy:

The repo panic was brought on by the Fed’s post-2008 contraction, which as the graph shows was forced into reverse. At the same time as the Fed was allowing its Treasury holdings to decline, the Treasury itself was issuing heavily, absorbing cash balances from other market participants. By late 2019, little cash was available for repo lending, and on September 17, intraday rates hit 10%, far above the Fed’s 2.25% upper limit for overnight money rates at the time, and far beyond the usual fluctuations in overnight markets.

The balance sheet expansion that followed seems modest in comparison to what was to come in 2020 and 2021, but it was a big deal at the time. The repo crunch had called into question the assumption that Treasury and repo markets are reliably liquid. One important consequence was the creation of the standing repo facility, which is designed to keep repo rates from spiking again. The SRF has yet to be activated, since pressure on rates has been downward not upward, but it seems likely to prevent a direct repeat of the 2019 crunch.

But other vulnerabilities have grown. In particular, hedge funds have continued to extend their short positions in Treasury futures, to nearly $1 trillion in notional open interest. The so-called basis trade involves pairing these futures positions with long positions in the cash Treasury market, funded with repo. Rising repo rates make the basis trade more costly, and if it were to be unwound in a disorderly way, it could certainly have wider repercussions for financial stability.

Previously on Soon Parted

• Day by day: The Fed in the Long Pandemic

• Getting back to normalization

• Cash–futures basis: The missing T accounts

Domestic production systems are connected by international trade and financial flows.

Up to now we have been considering a single unified production system, with all production grouped into a single monolithic “business” sector, and all spending grouped into monolithic “household” and “government” sectors. In the abstract, we could interpret this as a way of thinking of all human economic activity together, organized as a single global production system.

That would be a perfectly correct interpretation of what we have done so far, but macroeconomics more frequently focuses on national production systems, the economic activity carried out within some set of national borders. Why is national economy the usual choice of macroeconomic scope? One reason is that macroeconomic policy decisions are generally in the hands of government officials, and the sphere of influence of any particular government extends to its national borders. For example, stimulus spending is directed by government authorities, and monetary policy (for the most part) by national central banks. Similarly, most macroeconomic data is collected by government agencies, whose sphere of interest again extends to national borders and not beyond. (A notable exception is the supranational organization of European institutions, but this doesn’t really change the main point.)

So, if our three-sector model of production is interpreted as applying to a production system within the borders of one country, then we are missing something important: cross-border flows. As a first step toward adding these into our model, we can add a new sector called “rest of world,” the consolidation of all flows among countries other than the one we are interested in. What does this mean? Suppose we take the entire world’s economic activity, pull out one country of interest, China for example. Then we take all countries other than China and consolidate all of their economic flows into a single sector. Following the rules of consolidation, every flow will net out to zero, except those that cross into or out of China’s borders. The result might come out like this:

This four-sector model shows the three-sector model of a single domestic production system, plus one more sector for the rest of the world. Importantly, it doesn’t say anything about what is happening in the rest of the world, only what is crossing the borders of the country we are looking at. What crosses the borders? I’ve added two new kinds of flow, exports and imports. I imagine that the business sector does all of the foreign trade. Products that the domestic business sector buys from abroad, are imports for the business sector, a use of funds. Those same flows are exports for the rest of the world, a source of funds. Flows in the other direction work similarly.

However, I have made a strong assumption. Look at the sectoral account for the rest of the world: the only use of funds is imports, the only source is exports. By the rules of accounting, it must be that exports and imports are equal, so that the two entries balance correctly. But in the real world, this is clearly not the case: the US has run a persistent trade deficit (imports greater than exports) for many years; China has run a persistent trade surplus (exports greater than imports). One way to handle this is to lump imports and exports together as net exports, exports minus imports:

But if we do that, and we recognize that net exports can be something different from zero, then the rules of accounting immediately tell us that we need another entry. The rest of the world must have a source of funds, which must be a use of funds for one of the domestic sectors, and that source of funds must not be exports. Rather than leave you in suspense, I will tell you that this source of funds must be financial in nature, and in the modern world is frequently in the form of bonds. These must be bonds issued by the rest of the world and purchased by a domestic sector, which for now I assume to be the household sector.

How can we be sure that the household sector has the funds to buy these bonds? You can check this by adding numbers to the accounts, and using the rules of accounting to make the accounts balance. What you will see is that the flow of exports generates new incomes, which are paid from businesses to households. These higher incomes to households must be absorbed somewhere, and the shortest path is for households to buy foreign bonds. (There are other possibilities, but all of them end up with a domestic sector buying foreign bonds.)

When a country is a net exporter, exports greater than imports, net exports greater than zero, we say that the country is in current account surplus, and that country must also be a net purchaser of foreign financial assets.

When a country is a net importer, imports greater than exports, net exports less than zero, we say that the country is in current account deficit, and that country must also be a net issuer of financial assets held abroad.

Previously on Soon Parted

• Money is not yet tight

• The government sector: Macroeconomics, note 3

• Aggregation and consolidation: Macroeconomics, note 2

• Models: Macroeconomics, note 1

• Credit: Political economy of finance, note #1

• Sources and uses: A quiet plea for better accounting

Notes on charts, 1 May 2024

• The onshore CNY yield curve has flattened and may be inverting:

• So the FX forward curve is steepening:

• Forward markets in dollar-renminbi foreign exchange

• FX deposits at China’s banks have contracted by $250 billion

Short-term funding markets

• Money is not yet tight:

• Financial conditions beyond the fed funds rate

• Off the floor

• Along the curve: FRBNY sec lending

• A system under strain

• What’s going on with overnight rates?

• Quantitative tightening and overnight rates

• The overnight rate complex

Central bank balance sheets

Federal Reserve balance sheet

• Tax day has come and gone in the US:

• Here is the tax flow on US commercial banks’ balance sheet (reporting is one week behind the Fed):

• Fed’s balance sheet is still contracting. Following today’s FOMC announcement, contraction will be slower:

• ON RRP is below $500 billion:

• Money funds have moved from repo deposits to T-bills

• The debt ceiling, again

• Usage of the Fed’s ON RRP deposit facility has begun to contract

• Age of contraction

• The Fed’s balance sheet, 1914-1935

• Monetary impulse

• The overnight repo facility reaches $2 trillion

• The subtle sport of Fed-watching

• Mechanics of quantitative tightening: animating Zoltan

• Getting back to normalization

ECB balance sheet

• Contraction is proceeding also in the eurozone:

• The plumbing of monetary union

• Whatever it took

PBOC balance sheet

• This seems to be relatively normal seasonal contraction after the Lunar New Year:

Stablecoins

• At some point, holders of money-like instruments will surely begin to demand interest:

• Aldasoro, Mehrling and Neilson: A money view of stablecoins

The discourse on monetary policy is noisy. I can’t fix that unilaterally, but let me offer an observation: high interest rates do not necessarily mean that money is tight. I’ll go even further and say that right now, there are multiple reasons to think that money is not tight, and in fact that monetary conditions are still rather loose.

When is money tight?

When is money tight? Or in more sophisticated terms, when do monetary conditions restrain economic activity? It is common, both in academic discourse and in the financial press, to equate high interest rates with tight money. The argument, based on supply and demand, is familiar enough to seem intuitive: high interest rates mean borrowing is more expensive. This is thought to reduce demand for borrowing. Money is tight because rates are high.

Only modest reflection is needed to destabilize this intuition. The central objection is that people borrow in order to invest, and so the relevant consideration is not the cost of borrowing, but rather the difference between the cost of borrowing and the expected yield on the investment. If you can convince yourself that you can earn 7% yield, borrowing at 5% still seems cheap. The US is booming right now, evidenced by this week’s bumper retail sales figure for March, and headline unemployment still below 4%. Under such conditions, it is easy to rationalize high expected yields on investment. The technological novelty of large language models, for example, serves up such rationalizations on the daily.

Investment yields are analogous to interest income—the fixed interest income paid to a bond investor has much in common with the variable dividend income paid to an equity holder. More precisely, interest is not only an expense. It is an expense for borrowers, and ipso facto it is income for lenders and investors. High interest rates make life harder for some, but easier for others. So it need not be automatic that higher rates restrain economic activity, IS—LM very much notwithstanding.

The money markets provide what I think is the cleanest and highest-frequency way to measure this asymmetry. Below is a graph of daily dollar overnight interest rates. Rates are measured not as levels, but as spreads above the rate paid at the Fed’s overnight reverse repo (ON RRP) deposit facility. The upper and lower bounds on the fed funds rate, the rate paid on bank reserves, and the ON RRP rate itself are policy rates, thus straight lines; the other rates are market rates that fluctuate daily. The logic of this presentation is that holders of money can always get the ON RRP rate by depositing at the Fed, so under normal conditions they should never accept rates below that floor. To interpret the graph, one need not dwell on the precise readings of each rate; just look at where the whole complex of rates sits relative to zero. With that in mind, the graph tells a compelling story:

In 2020 and 2021, the Fed flooded the financial system with newly created reserves, the liability counterpart to its almost $5 trillion in asset purchases. Overnight rates were driven down toward the ON RRP rate floor, and some rates frequently leaked below the floor, mostly in 2022. In other words, money was loose, so loose that plenty of lenders would accept less interest than what the Fed was paying. Since 2023, the overnight rate complex has drifted back up, mostly off the floor and much more rarely below it. This is connected to the increases in interest rates, marked by vertical lines in the graph. But it is not identical to those increases—rates have jumped up while spreads have drifted, mostly up but not only up.

And now?

And now? Viewed through overnight rates, money is certainly tighter than in 2023. But some of these rates still print at or below ON RRP with some frequency, and have only occasionally tested the upper half of the Fed’s range. Market power, by this measure, still lies more on the side of borrowers than on that of lenders.

What does this mean for the Fed and for Fed-watchers? The US central bank’s problem is that its main instrument, the fed funds rate, does not exert a direct influence on its main policy target, PCE inflation. The connection is weak, and other factors can intervene. And as the graph above shows, higher rates do not make money tighter if lenders are still fighting to lend.

Fed-watchers would do better, in my opinion, to focus on balance-sheet contraction, the Fed’s ongoing process of allowing its asset portfolio to run off. This has taken a lot of money off the table, and sooner or later will lead lenders to tighten their fists.

Previously on Soon Parted

• Models: Macroeconomics, note 1

• The basis trade and repo rates

• Money funds have moved from repo deposits to T-bills

• Off the floor

Short-term funding markets

• Money funds have moved from repo deposits to T-bills

• The debt ceiling, again

• Money is valuable again: Financial conditions beyond the fed funds rate

• Off the floor

• Along the curve: FRBNY sec lending

• A system under strain

• What’s going on with overnight rates?

• Quantitative tightening and overnight rates

• The overnight rate complex

Central bank balance sheets

Federal Reserve balance sheet

• Usage of the Fed’s ON RRP deposit facility has begun to contract

• Age of contraction

• The Fed’s balance sheet, 1914-1935

• Monetary impulse

• The overnight repo facility reaches $2 trillion

• The subtle sport of Fed-watching

• Mechanics of quantitative tightening: animating Zoltan

• Getting back to normalization

ECB balance sheet

• The plumbing of monetary union

• Whatever it took

BOJ balance sheet

• Yield curve control

• The Bank of Japan

SNB balance sheet

• The SNB taps its swap line

PBOC balance sheet

• The PBOC is buying foreign exchange

• China’s difficult choices: the missing T accounts

• China in the global production system

• The US–China clearing balance

• The PBOC balance sheet, part 2

• The PBOC standing bid

• Forward markets in dollar-renminbi foreign exchange

• FX deposits at China’s banks have contracted by $250 billion

Stablecoins

• Aldasoro, Mehrling and Neilson: A money view of stablecoins

US commercial banks

An important debate in the history of macroeconomics tries to resolve this important question: “In a market economy, can a government redirect spending using taxes and borrowing, so as to increase total incomes?” This question was of central importance in the global depression of 1929-1933: around the world, total spending, and so also household incomes, dropped sharply. The consequence was high unemployment and so decreased access to the means of human subsistence.

Some economists argued that it might be that, using the centralized spending power of a government, it would be possible to change the patterns of spending, increasing incomes and improving people’s lives. In other words, it might be possible using centralized economic decision-making to improve upon the results of decentralized market-driven decisions. This argument is strongly associated with the name of John Maynard Keynes, an English economist whose work during and after the Great Depression helped create macroeconomics, and changed the way that we think about recessions and depressions. We should be clear that this question is inherently political-economic. By changing how decisions are made about resource allocation, necessarily some people will end up with more resources and others will end up with less. The politics are important, but let’s first try to model the spending flows.

The main theoretical task is to fit a government sector into our picture of macroeconomic flows. In both the two-sector model of production and in the supply-chain model, we had only households and businesses, with no explicit place for production. But in most places today, governments control a significant fraction of economic flows, and they decide how to direct those flows in substantially different ways than either households or businesses. So the added complexity of an additional sector seems like a fair price to pay for capturing the important macroeconomic impact of government.

This three-sector model is a step in that direction. Previously we had used the label “product” to refer to businesses’ source of funds; now we divide that into two categories, household consumption and government spending. In macroeconomic equations, these are usually written as C and G respectively. Both are flows, monetary values associated with a duration in time, as are all the entries in this model. In words, we still have a single sector in which all production takes place, “business.” This business sector sells its output to households as before, but now also to government. You can think of a government contracting with a private business to build a road or an airport.

In simpler models, we defined gross domestic product as the total of all production flows that are sources of funds to business. That is still the case, but now there are two components to that production. So GDP in this model is consumption plus government spending, Y = C + G.

This model also includes a schematic version of government finance. The government sector has sources of funds of two kinds, taxation and bond issuance. Taxes are income flows, legitimated through the legal system and collected from households and businesses under threat of punishment. Tax is a one-time flow, an expense to households and businesses that is gone once the tax has been paid. Governments can also fund themselves by issuing bonds, which is a way of borrowing money. Bonds are different from taxes, because they entail a future flow in the opposite direction, when the bonds are repaid. Repayment is not shown in our T accounts—for now we are thinking only about new borrowing. Note that I write a plus sign, “+Bonds,” which indicates that a stock of outstanding bonds continues to exist after this period is done, unlike taxes, incomes, consumption or government spending.

We can use this model to consider a couple of scenarios. Suppose that the economic policymakers in a government observe that household incomes are low, so that people’s access to material necessities and comforts is limited. Such a government might choose to increase government spending as a stimulus measure. Increasing government spending adds to businesses’ sources of funds, so increasing also the incomes paid to households.

But to keep our model in balance, increased spending by government must be paid for. If it does so by increasing taxes on business, then you can see that the increased tax burden will completely absorb the new spending flows, and household incomes won’t rise at all. If government pays for the increased spending by increasing taxes on households, then incomes will rise, but the entire increase will be absorbed by the higher taxes. Or if government pays for the increased spending by issuing new bonds, then it will need to sell those bonds, again absorbing any increased income to households.

But this doesn’t mean that stimulus spending can’t increase incomes. In fact, there are a couple of ways that government can increase total incomes. We can see one of them in the model we have so far. If government can use its spending power to convince households that incomes are improving, then households might themselves choose to spend more. How do they afford the increased spending? Interestingly, consumption spending is a source of funds to businesses, who therefore pay out more incomes to households: the increased spending pays for itself.

Previously on Soon Parted

• Aggregation and consolidation: Macroeconomics, note 2

• Models: Macroeconomics, note 1

• The PBOC is doing something big

• Credit: Political economy of finance, note #1

• Sources and uses: A quiet plea for better accounting

• The Fed’s debt ceiling memo

Central bank balance sheets

Federal Reserve balance sheet

• Usage of the Fed’s ON RRP deposit facility has begun to contract

• Age of contraction

• The Fed’s balance sheet, 1914-1935

• Monetary impulse

• The overnight repo facility reaches $2 trillion

• The subtle sport of Fed-watching

• Mechanics of quantitative tightening: animating Zoltan

• Getting back to normalization

ECB balance sheet

• The plumbing of monetary union

• Whatever it took

PBOC balance sheet

• The PBOC is doing something big

• The PBOC is buying foreign exchange

• China’s difficult choices: the missing T accounts

• China in the global production system

• The US–China clearing balance

• The PBOC balance sheet, part 2

• The PBOC standing bid

Short-term funding markets

• Money funds have moved from repo deposits to T-bills

• The debt ceiling, again

• Money is valuable again: Financial conditions beyond the fed funds rate

• Off the floor

• Along the curve: FRBNY sec lending

• A system under strain

• What’s going on with overnight rates?

• Quantitative tightening and overnight rates

• The overnight rate complex

Global dollar

• Seller’s market

• Forward markets in dollar-renminbi foreign exchange

• FX deposits at China’s banks have contracted by $250 billion

Let’s look again at the production sector. Up until now we have used a two-sector economic model, in which all production takes place in a single “business” sector. We can make this a little more elaborate, which will let us speak to some important macroeconomic questions. We’ll do this using a macroeconomic model in which production takes place across three sectors, the final output of which is sold to a single household sector. Production across the sectors is linked in a very particular way, which we can describe as a supply chain.

The supply chain model is written out using sectoral T accounts below. Households are on the left. The business sectors are arranged from right to left. The first business sector we refer to as extractors—a representative example would be a business that mines copper from the earth, or a farm that produces ethanol from corn. We think of the output of the extractive sector as providing inputs to manufacturers, who combine these resources into finished goods. Finally, retailers sell these finished goods to households:

The arrangement of T accounts follows the examples we have seen before. Households earn incomes, their sources of funds, and spend these incomes buying product from the retail sector. What is different is that this model incorporates intermediate product, the sales of product from one business sector to another, on their way to eventual final sale. Manufacturers, for example, take in the product of the extractive sector, a use of funds for manufacturers and a source of funds for extractors. They then sell their products to retailers, a use of funds for retailers and a source of funds for manufacturers.

It will be helpful to see the relationship between this supply-chain macroeconomic model and the two-sector model of production that we have seen before. We can see the connection in two steps. First, write the transactions for the three business sectors (extraction, manufacturing and retail) on a new T account for a single business sector. This is called aggregation, and it means that we take what had been multiple different sectors and group them to think of them as a single sector. The resulting model looks like this:

This creates an interesting complication, which is new to us with this example. After aggregation, our business sector shows several transactions with itself. In a way, the rules of accounting are still satisfied, since every source has a use, and total sources equal total uses for each sector. But now we have the aggregated business sector transacting with itself—the product of extractors and the product of manufacturers stay within the business sector. We can remove them in a step called consolidation, removing transactions that stay within a sector:

In other words, by aggregating and consolidating the supply-chain model, we come back to the original two-sector macroeconomic model.

Previously on Soon Parted

• Models: Macroeconomics, note 1

• The PBOC is doing something big

• Credit: Political economy of finance, note #1

• Sources and uses: A quiet plea for better accounting

• The Fed’s debt ceiling memo

In macroeconomics, as in economics more generally, one frequently talks about a conceptual “model” of an economic system. But what do we mean by this word model? To start, you can picture a model airplane, or a model of the Solar System. This kind of model shows the largest components of the thing being modeled, and represents the relationships between those parts. However, the model is smaller, simpler and less complete than the original. An economic model is this kind of thing. It tries to illustrate the most important parts of an economic system in a context that is smaller and easier to understand than the real thing.

Our two-sector division of the macroeconomy is a first model of the system of production, labor and exchange. Below is one way we can write out this two-sector model. In this style of writing out macroeconomic relationships, each sector gets a two-column T account. At the moment, we are working with two sectors, a household sector and a business sector, but later we will have more. The two columns of each sector are labeled “uses” and “sources.” We will always write uses on the left, and sources on the right. That’s just a convention—the only reason to do it that way is so that we can all understand each other.

Sources and uses are short for sources of funds and uses of funds. What does this mean? We think of each sector as receiving payments, which are its sources of funds, and making payments, which are its uses of funds. The household sector, for example, receives payments in the form of incomes from business. That includes the regular wages and salaries that workers receive as compensation for their labor. It also includes profits paid to the owners of businesses, as well as interest and dividends on savings and investments. In other words, we are grouping together all household incomes. This is a source of funds to the household sector, and we write it in the right-hand column, labeled “sources,” for that sector. These same income payments are a use of funds for the business sector, so we write it in the left-hand column, labeled “uses,” for that sector:

Households also make payments. In this simple model, the only thing households can do with their funds is to buy the product of businesses. We write that on the left-hand side for Households, under uses of funds. Those same payments become a source of funds for the business sector.

In a macroeconomic model, we imagine payments coming in and payments going out for each sector. We do not allow payments to appear from nowhere, nor do we allow them to disappear. Every payment coming in to a sector has to go somewhere, and every payment going out has to come from somewhere. We formalize this with two important rules. First, for each sector, total sources of funds must equal total uses of funds. If for some sector total sources and total uses are not equal, that means we have missed something and our model is incomplete. In the two-sector model we are looking at right now, household sources of funds and household uses of funds are both $27 trillion. Business sources and uses also total $27 trillion. So the first rule is satisfied.

Second, each entry represents funds flowing from a sector to another sector. Funds are not allowed to appear or disappear between sectors. So, each entry appears once as a source of funds and once as a use of funds, in equal amounts. Above, we show incomes as a source of funds for households and a use of funds, in the same amount, for businesses. We show sales of product as a source of funds for businesses, and a use of funds for households. So the second rule is also satisfied.

To be clear, the two rules are:

1. total uses equal total sources for each sector, and

2. each entry appears as a use and a source, in equal amounts.

In order to make progress in macroeconomics, we will need to be able to think clearly about sectoral relationships. This way of writing them down, with T accounts and sources and uses accounting, is a good way to do it, because by following the two rules you will avoid some mistakes that would otherwise be easy to make.

Finally, we should ask whether this model is a good one. It is pretty simple, reducing a year’s worth of production to two two-sector entries. It is meaningful, because the household-businesss sector does give rise to reasonable definition of GDP and this is the most common measure of total economic activity in an area during a period of time. At the same time, it seems to have some obvious omissions—there is only one country for example, and only one kind of business, no government, and so on. We will quickly need to move on to more elaborate versions.

Previously on Soon Parted

• The PBOC is doing something big

• Credit: Political economy of finance, note #1

• Sources and uses: A quiet plea for better accounting

• The Fed’s debt ceiling memo

Central bank balance sheets

Federal Reserve balance sheet

Read more

One of the asset entries, loans from the PBOC to Chinese commercial banks, has exploded, picking up 5 trillion renminbi in the last six months. The central bank is lending money at a furious pace. One can compare to 2016, when PBOC loans to commercial banks spiked by a similar amount. But at that time the central bank was selling assets too, leaving the balance sheet mostly level overall:

Now, in significant contrast, we are seeing outright balance sheet expansion, quantitative easing of some kind, on a scale that is notably changing the structure of the PBOC’s balance sheet. At this pace, for example, it won’t be long before the PBOC holds more Chinese commercial bank credit than it does Treasury securities.

The lending has been balanced on the liability side, first by expansion in government deposits (which I record under other liabilities), which then contracted and were replaced by a big jump in bank reserves.

These PBOC balance sheet developments seem worth noting. Unfortunately I don’t have a clear story to offer just yet. Please point me toward any sources that clarify what the PBOC is trying to do.

Stablecoins paper

• What drives stablecoin issuance?

• Forward Guidance: Stablecoins Are The New Eurodollars

• Aldasoro, Mehrling and Neilson: A money view of stablecoins

• Stablecoins paper PDF

Previously on Soon Parted

• Forward markets in dollar-renminbi foreign exchange

• FX deposits at China’s banks have contracted by $250 billion

• FX reserves of Chinese banks

• Foreign assets in the Chinese banking system

Stablecoin issuance rose through 2021, then peaked and fell during 2022 and 2023. In the paper, we measured the returns to holding stablecoins using lending rates in the Aave DeFi lending protocol. Aave is relatively conservative among DeFi protocols; accordingly it pays less than other protocols, so this spread should a lower bound to DeFi yields. When off-chain interest rates were zero, as they were until early 2022, one could borrow money for free, buy stablecoins, and lend them out at a positive rate using Aave or another protocol.

The profitability of the entire strategy can measured by the spread of Aave lending rates over SOFR, a key off-chain overnight rate. This graph captures the story. It is a version of Graph 3 in the paper, presented here using Soon Parted style:

A simple and compelling story emerges: investment in stablecoins grew while defi protocols offered high returns; when those returns turned negative, investment in stablecoins fell.

We finished the paper at the beginning of November. Since then, Aave lending rates have risen, so that the spread turned positive once again. As the graph shows, stablecoin issuance stopped declining and began to rise. This seems like a very nice out-of-sample confirmation.

Please do check out the paper for the full analysis.

Stablecoins paper

• Forward Guidance: Stablecoins Are The New Eurodollars

• Aldasoro, Mehrling and Neilson: A money view of stablecoins

• Stablecoins paper PDF

Previously on Soon Parted

• The basis trade and repo rates

• Cash–futures basis: The missing T accounts

• The Fed’s debt ceiling memo

The same CFTC data shows that asset managers are taking the corresponding long positions:

Why? Over the last two years, rates have been rising, meaning that bond prices have been falling. Asset managers, sensitive to these falling prices, enter long futures contracts on Treasuries to shorten the durations of their portfolios, so reducing their exposure to interest-rate risk. These purchases drive futures prices up relative to spot prices, or in other words they cause the cash–futures basis to increase.

Hedge funds see this widening basis and enter the offsetting short positions in Treasury futures, borrowing the funds from their dealers using repo. Dealers in turn borrow from money market funds, also using repo. A recent piece in the BIS Quarterly Review, by my co-author Iñaki Aldasoro and his colleague at the BIS Sebastian Doerr, notes that dealers are using the DTCC’s sponsored repo service for these trades, to minimize the amount of capital they need to post.

Intriguingly, repo rates on the FICC’s DVP and GCF repo services have spiked relative to other overnight rates, pulling the benchmark SOFR average up with them:

Hedge funds have to pay those rates in order to stay in the Treasury basis trade. If they stay high, expect the basis trade to unwind more quickly.

Previously on Soon Parted

• Cash–futures basis: The missing T accounts

• The Fed’s debt ceiling memo

• Money is valuable again: Financial conditions beyond the fed funds rate

Stablecoins paper

• Forward Guidance: Stablecoins Are The New Eurodollars

• Aldasoro, Mehrling and Neilson: A money view of stablecoins

• Stablecoins paper PDF: https://www.bis.org/publ/work1146.htm